Fix nccl for remote builds.

Instead of symlinking the install dir, copy the two files we need.
Symlinking a system dir like /usr is generally problematic as it can quickly
lead to miscompiles for unrelated reasons. Furthermore, bazel will consider
it an error if /usr is linked in and contains a recursive symlink in
/usr/bin/X11 -> .

PiperOrigin-RevId: 211842260

third_party/nccl/nccl_configure.bzl

@@ -16,6 +16,7 @@ load(
 
 _NCCL_INSTALL_PATH = "NCCL_INSTALL_PATH"
 _TF_NCCL_VERSION = "TF_NCCL_VERSION"
+_TF_NCCL_CONFIG_REPO = "TF_NCCL_CONFIG_REPO"
 
 _DEFINE_NCCL_MAJOR = "#define NCCL_MAJOR"
 _DEFINE_NCCL_MINOR = "#define NCCL_MINOR"
@@ -48,25 +49,8 @@ alias(
 """
 
 # Local build results in dynamic link and the license should not be included.
-_NCCL_LOCAL_BUILD_TEMPLATE = """
-filegroup(
-  name = "LICENSE",
-  visibility = ["//visibility:public"],
-)
-
-cc_library(
-  name = "nccl",
-  srcs = ["nccl/lib/libnccl.so.%s"],
-  hdrs = ["nccl/include/nccl.h"],
-  include_prefix = "third_party/nccl",
-  strip_include_prefix = "nccl/include",
-  deps = [
-      "@local_config_cuda//cuda:cuda_headers",
-  ],
-  visibility = ["//visibility:public"],
-)
-"""
-
+_NCCL_REMOTE_BUILD_TEMPLATE = Label("//third_party/nccl:remote.BUILD.tpl")
+_NCCL_LOCAL_BUILD_TEMPLATE = Label("//third_party/nccl:system.BUILD.tpl")
 
 def _find_nccl_header(repository_ctx, nccl_install_path):
   """Finds the NCCL header on the system.
@@ -137,6 +121,13 @@ def _nccl_configure_impl(repository_ctx):
     repository_ctx.file("BUILD", _NCCL_DUMMY_BUILD_CONTENT)
     return
 
+  if _TF_NCCL_CONFIG_REPO in repository_ctx.os.environ:
+    # Forward to the pre-configured remote repository.
+    repository_ctx.template("BUILD", _NCCL_REMOTE_BUILD_TEMPLATE, {
+        "%{target}": repository_ctx.os.environ[_TF_NCCL_CONFIG_REPO],
+    })
+    return
+
   nccl_version = repository_ctx.os.environ[_TF_NCCL_VERSION].strip()
   if matches_version("1", nccl_version):
     # Alias to GitHub target from @nccl_archive.
@@ -148,8 +139,10 @@ def _nccl_configure_impl(repository_ctx):
     # Create target for locally installed NCCL.
     nccl_install_path = repository_ctx.os.environ[_NCCL_INSTALL_PATH].strip()
     _check_nccl_version(repository_ctx, nccl_install_path, nccl_version)
-    repository_ctx.symlink(nccl_install_path, "nccl")
-    repository_ctx.file("BUILD", _NCCL_LOCAL_BUILD_TEMPLATE % nccl_version)
+    repository_ctx.template("BUILD", _NCCL_LOCAL_BUILD_TEMPLATE, {
+        "%{version}": nccl_version,
+        "%{install_path}": nccl_install_path,
+    })
 
 
 nccl_configure = repository_rule(

third_party/nccl/remote.BUILD.tpl

@@ -0,0 +1,6 @@
+licenses(["restricted"])
+
+package(default_visibility = ["//visibility:public"])
+
+alias(name="LICENSE", actual = "%{target}:LICENSE")
+alias(name = "nccl", actual = "%{target}:nccl")

third_party/nccl/system.BUILD.tpl

@@ -0,0 +1,26 @@
+filegroup(
+  name = "LICENSE",
+  visibility = ["//visibility:public"],
+)
+
+cc_library(
+  name = "nccl",
+  srcs = ["libnccl.so.%{version}"],
+  hdrs = ["nccl.h"],
+  include_prefix = "third_party/nccl",
+  deps = [
+      "@local_config_cuda//cuda:cuda_headers",
+  ],
+  visibility = ["//visibility:public"],
+)
+
+genrule(
+  name = "nccl-files",
+  outs = [
+    "libnccl.so.%{version}",
+    "nccl.h",
+  ],
+  cmd = """cp "%{install_path}/include/nccl.h" "$(@D)/nccl.h" &&
+           cp "%{install_path}/lib/libnccl.so.%{version}" "$(@D)/libnccl.so.%{version}" """,
+)
+